Recherche avancée

Sur d’autres sites (4897)

• Conversion of mp3 to flac results in file with longer duration

  18 juin 2019, par ThaDon

  I’ve noticed that when I convert an mp3 file to flac, the duration reported in the flac file will often differ from that of the source mp3 file. Mostly this difference is negligible and can be ignored (perhaps a fraction of a second).

  However, there are times when the timing is off by several seconds, and this causes my processing pipeline quite a bit of problem.

  For instance, take this podcast episode for example. If I run it through ffmpeg, I can see that it has a duration of :

  Duration: 00:52:38.39, start: 0.000000, bitrate: 128 kb/s

  If I then convert it to flac using the following command :

  ffmpeg -i startups-for-the-rest-of-us-448.mp3 -ac 1 -ar 16000 -f flac output.flac

  I can see that the duration of the flac file is :

  Duration: 00:52:45.65, start: 0.000000, bitrate: 133 kb/s

  Note there is an error message during conversion that states :

  [mp3 @ 0x7fffd16d6780] Header missing
  
  Error while decoding stream #0:0: Invalid data found when processing input

  Does the difference in duration have have to do with the bitrate difference ? When I listen to the file it sounds identical, I’m assuming the flac version must be ever so slightly slower as to gain the extra 7 seconds over the course of the podcast.

• Concatenation and Transcoding of MPEG-TS files FFMPEG

  20 décembre 2020, par SquawkBirdies

  I have a DVR which records over-the-air TV and saves the recordings in the MPEG-TS format, splitting each episode across multiple files, each 500 MB in size.

  



  To simplify archiving, I have been trying to write a shell script to automate the process of concatenating the files together and transcoding them into a more common video format, like h.264.

  



  Here are the steps I have performed so far :

  



  

  • I wanted to make sure that the files I was getting were valid in the first place. To test this, each section was transcoded in Handbrake before being merged using ffmpeg's concat command. This worked, but was manual and added an annoying black frame between each section.

  


  • I wrote a shell script to find all the sections of an episode in a folder and put the file names into a text file that the concat demuxer could parse.

  


  • Tested this command :

  


  



  


  


  ffmpeg -hide_banner -f concat -i video_files_tmp.txt -c:v libx264 -pix_fmt yuv420p -c:a aac $2$OUTPUT_FILE_NAME

  


  


  




  During the transcode, this would throw many warnings and errors, such as "PES packet size mismatch". At some point, it would warn that more than 1,000 frame were skipped. When the output was played, it would skip frames and result in a file where the video froze partway through but the audio continued playing. I tried adding -vsync 0 to the output as well.

  



  

  • Then, tried splitting the concatenation and transcode into two steps :

  


  



  


  


  ffmpeg -hide_banner -f concat -i video_files_tmp.txt -c copy output_tmp.ts
ffmpeg -hide_banner -i output_tmp.ts -c:v libx265 -pix_fmt yuv420p -c:a aac $2$OUTPUT_FILE_NAME

  


  


  




  This did basically the same thing as before.

  



  

  • Tried using the libx265 encoder instead. Same result.

  


  • Then, I tried just playing the concatenated MPEG-TS file directly, which also would freeze at some point during the playback.

  


  



  I was wondering about what ffmpeg flags or options to set to get this working, or other options I could try ? Thanks !

  


• Kodi : playback several video files from the internet as single movie with single timeline

  30 mai 2018, par HarryFox

  I’m writing an add-on for Kodi in Python, for a site with movie collection, which can be able to playback online video from the site. But problem is that many of them splitted apart on pieces of different length. And its not mpeg dash nor m3u playlists, its just individual pieces of single movie or an episode.

  My target is able to play those splitted videos as single movie with single timeline and playback will no require download all pieces of a movie. It’s important because many of Movies\episodes provides with subtitle file, that why single timeline is important. Also general idea for plugin which I’m writing is comfort to use, so online playback is very important too.

  I did some research but it almost no result. On official Kodi forum advise to create Input Stream add-on like Input Stream Adaptive, but it was written on c++ and it’s unattainable for me (for now at least). There is no tools for python to create such kind of add-ons.

  Another idea is to create middle server which on fly will somehow (with ffmpeg) combine those pieces, but it seems that the process will too resource-intensive especially for TV-boxes.
  In this case also raise many question about which way is better, and i have no enough experience and knowledge to solve the problems by myself even with google.

  So i need an advice how can i solve the problem, just to know right direction.
  Thank you for your time.
  Sorry for my English.